INDEX

Explanations

how questions

np_acts-logits-general · gemini-2.5-flash-lite

The neuron activates on question-introducing words—especially “How” (and its variants like “How much/many/…”)—marking interrogative sentence starts.

oai_token-act-pair · o4-mini Triggered by @jyhe0408

interrogative phrases beginning with "How" in questions.

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

questions framed with interrogatives, especially those asking about manner, degree, quantity, or time.

oai_token-act-pair · gpt-5 Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 அது

0.65

 privileges

0.64

 fixtures

0.63

 diodes

0.61

 generalizations

0.60

 predom

0.59

 molded

0.59

äne

0.58

அது

0.58

 perfusion

0.57

POSITIVE LOGITS

Much

1.28

 Much

1.27

 much

1.18

 MUCH

1.16

much

1.14

 does

0.86

 mucho

0.86

did

0.82

 Many

0.82

Many

0.81

Activations Density 0.212%

how questions

The neuron activates on question-introducing words—especially “How” (and its variants like “How much/many/…”)—marking interrogative sentence starts.

interrogative phrases beginning with "How" in questions.

questions framed with interrogatives, especially those asking about manner, degree, quantity, or time.

No Comments

No Known Activations